state-space observers 4 system stabilitycontroleducation.group.shef.ac.uk/statespace/state...
TRANSCRIPT
State-space observers 4 system stability
J A Rossiter
1
Slides by Anthony Rossiter
Introduction
The previous video demonstrated that one could derive an observer gain L to give the desired state estimation error dynamics.
However, in practice an observer gives state estimates used by state feedback for system control. What is the impact on overall stability?
Slides by Anthony Rossiter
2
Czy
yyLBuAzz
m
m )(
))(( zxLCAzx
Cxy
BuAxx
Overall structure
Combining the observer, state feedback and system dynamics requires a number of parallel models.
Is the overall system of equations stable?
Slides by Anthony Rossiter
3
Czy
yyLBuAzz
m
m )(
Kzu
Cxy
BuAxx
For now, assume no parameter uncertainty.
Combine equations
The easiest way to analyse the coupled dynamic equations is to construct a single equivalent state space model with states x and z.
Slides by Anthony Rossiter
4
CzyyyLBuAzz
CxyKzuBuAxx
mm );(
,;
z
x
LCBKALC
BKA
z
x
Next, use a similarity transform to help expose the underlying modes.
Similarity transformation
One can easily expose the underlying modes.
Slides by Anthony Rossiter
5
z
x
LCBKALC
BKA
z
x
Transformed matrix is now block diagonal
LCA
BKBKAT
LCBKALC
BKAT
II
IT
II
IT
0
;0
;0
1
1
Overall modes Key modes are taken from the eigenvalues of the transformation matrix.
Slides by Anthony Rossiter
6
Closed-loop poles when combining state feedback with an observer are the poles from the state
feedback design combined with the poles from the observer design.
0
00
0
LCAI
BKAIAI
LCA
BKBKAA
c
c
Remark – separation principle
One can design the state feedback and observers independently one of an other without any impact on overall closed-loop stability.
NOTE however:
• We have assumed no parameter uncertainty.
• We have not considered the impact of the combined design on overall behaviour.
Slides by Anthony Rossiter
7
MATLAB ILLUSTRATIONS
We will compare performance with an observer and assuming states are measureable.
How does overall behaviour change with different choices of observer poles?
We will not look at the impact of parameter uncertainty.
Slides by Anthony Rossiter
8
Make observer poles 10x faster than system poles so that state errors
converge quickly
Slides by Anthony Rossiter
9
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2-70
-60
-50
-40
-30
-20
-10
0
10
20
e with observer
y with observer
y without observer
input with observer
input without observer
Transient input very
large
Fast error convergence
Make observer poles 2x faster than system poles so that state errors
converge moderately quickly
Slides by Anthony Rossiter
10
Transient input still
large
Fast error convergence
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2-15
-10
-5
0
5
10
e with observer
y with observer
y without observer
input with observer
input without observer
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2-1
0
1
2
3
4
5
6
7
8
9
e with observer
y with observer
y without observer
input with observer
input without observer
Make observer poles slower than system poles.
Slides by Anthony Rossiter
11
Transient input small as observer state slow to move.
Poor error convergence
Behaviour in effect open-loop!
Conclusion Introduced the separation principle used for analysing system stability with state feedback and a state observer. 1. Shown that one can independently design the state
feedback and observer gain to give specified poles and these poles are inherited by the overall closed-loop system, in the absence of parameter uncertainty.
2. Nevertheless, stability and performance are different things and it is less straightforward to determine the impact of observer pole selection on overall closed-loop behaviour.
3. Consideration of these issues and others is beyond the current remit of this video series. Slides by Anthony Rossiter
12
© 2016 University of Sheffield This work is licensed under the Creative Commons Attribution 2.0 UK: England & Wales Licence. To view a copy of this licence, visit http://creativecommons.org/licenses/by/2.0/uk/ or send a letter to: Creative Commons, 171 Second Street, Suite 300, San Francisco, California 94105, USA. It should be noted that some of the materials contained within this resource are subject to third party rights and any copyright notices must remain with these materials in the event of reuse or repurposing. If there are third party images within the resource please do not remove or alter any of the copyright notices or website details shown below the image. (Please list details of the third party rights contained within this work. If you include your institutions logo on the cover please include reference to the fact that it is a trade mark and all copyright in that image is reserved.)
Anthony Rossiter Department of Automatic Control and
Systems Engineering University of Sheffield www.shef.ac.uk/acse